Adds comment explaination of when to resample

README.md

@@ -102,6 +102,8 @@ model = Wav2Vec2ForCTC.from_pretrained("facebook/wav2vec2-large-960h-lv60-self")
 processor = Wav2Vec2Processor.from_pretrained("facebook/wav2vec2-large-960h-lv60-self")
 
 def map_to_pred(batch):
+    # LibriSpeech sampling rate (16kHz) is equal to Wav2Vec2 processor sampling rate -> pass audio directly to processor
+    # Note that in case the sampling rates do not match you would have to resample!
     inputs = processor(batch["audio"]["array"], return_tensors="pt", padding="longest", sampling_rate=batch["audio"]["sampling_rate"])
     input_values = inputs.input_values.to("cuda")
     attention_mask = inputs.attention_mask.to("cuda")